AI’s First Big Shock to the Job Market

FEATURED INSIGHT💡

Stanford Study Reveals AI’s First Big Hit on Jobs

A new study from Stanford economists Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen offers the clearest evidence yet that generative AI is reshaping the labor market, and the first to feel the impact are young, entry-level workers.

Using payroll data from over 5 million U.S. workers through July 2025, the researchers tracked employment across occupations with different levels of AI exposure. Their findings suggest that AI’s impact is not uniform but highly concentrated:

• Early-career decline: Workers aged 22–25 in the most AI-exposed jobs (such as software developers and customer service reps) saw a 13% relative drop in employment since late 2022, even after controlling for firm-level shocks. Older colleagues in the same roles remained stable, or even grew in employment.

• Automation vs. augmentation: Jobs where AI is used mainly to automate tasks show clear employment losses for younger workers. But in roles where AI primarily augments human work, entry-level hiring has actually held steady or grown.

• Wages hold steady (for now): Paychecks haven’t shifted much. The adjustment is happening through employment, not compensation, suggesting wage stickiness in the short run.

• Not just tech: While the effect is most visible in software and customer service, the pattern persists across a range of occupations, even after excluding tech firms and remote-friendly jobs.

Why are younger workers hit harder? The authors argue that AI replaces “codified knowledge” (the book-learning that new grads bring to the job) more easily than the “tacit knowledge” accumulated through years of experience. In other words, AI is best at doing the things schools teach, but not the subtle skills you only pick up on the job.

Big Picture: Overall employment in the U.S. is still growing, but the study suggests a looming risk: the entry-level career ladder may be eroding in AI-exposed fields. This could have ripple effects on how the next generation of workers builds skills, gains experience, and advances over time.

Create How-to Videos in Seconds with AI

Stop wasting time on repetitive explanations. Guidde’s AI creates stunning video guides in seconds—11x faster.

• Turn boring docs into visual masterpieces

• Save hours with AI-powered automation

• Share or embed your guide anywhere

How it works: Click capture on the browser extension, and Guidde auto-generates step-by-step video guides with visuals, voiceover, and a call to action.

Best part? It’s 100% free.

👉 Download the extension now

ON THE HORIZON 🌅

The U.S. Government Buys Into Intel

In a move with echoes of the 2008 auto bailout, the White House is converting $8.9B in Chips Act grants into equity, taking a 10% stake in Intel. That makes the government the company’s third-largest shareholder, right behind BlackRock and Vanguard.

The deal aims to revive domestic chipmaking and secure U.S. control over critical tech supply chains. But it also marks a philosophical shift: Washington is no longer just funding industry, it’s owning a piece of it.

Intel’s stock jumped 6% on the news, though questions remain about governance and long-term competitiveness.

LATEST IMPORTANT NEWS 📰

Meta + Midjourney: Visual AI at Scale

Meta struck a deal to license Midjourney’s “aesthetic technology” for its future AI products. Expect to see Midjourney-powered visuals baked into Instagram, WhatsApp, and Facebook before long.

Anthropic’s Data Dilemma

Claude users now face a new choice: opt out—or let Anthropic train on your conversations for five years. Business customers are exempt, but consumer users (Free, Pro, Max, Claude Code) must actively toggle off data sharing. Critics say the design nudges people into consent by default.

Nvidia’s $46.7B Quarter

Nvidia just posted record revenue up 56% YoY, powered by demand for its Blackwell chips. Net income soared to $26.4B. CEO Jensen Huang called AI infrastructure a “$3–4 trillion opportunity” over the next five years. Meanwhile, China sales remain tangled in export restrictions and politics.

FOR THE TECHNICALLY INCLINED 🛠️

Fine-Tuning Agents Without Touching the Models

A new paper, Memento: Fine-tuning LLM Agents without Fine-tuning LLMs, introduces a way to make AI agents smarter without retraining the base models.

Instead of expensive gradient updates, Memento uses memory-based reinforcement learning. Agents store experiences in an episodic memory and adapt on the fly by rewriting and retrieving from it.

The result? State-of-the-art performance on deep research tasks, with top scores on the GAIA and DeepResearcher benchmarks. It’s a scalable path to real-time, continuous learning agents, without the GPU burn of traditional fine-tuning.

100 Genius Side Hustle Ideas

Don't wait. Sign up for The Hustle to unlock our side hustle database. Unlike generic "start a blog" advice, we've curated 100 actual business ideas with real earning potential, startup costs, and time requirements. Join 1.5M professionals getting smarter about business daily and launch your next money-making venture.

Get the guide

AI TOOL OF THE DAY 🚀

Rocket – Build an App From One Prompt. Type a single prompt, and Rocket spits out a production-ready app—mobile, web, or dashboard. It’s fast, secure (paid plans don’t train on your data), and comes with Figma-to-app baked in. Great for founders, agencies, and teams that need to ship quickly.